Re: extend diff3 to ignore whitespace differences - here's a patch

[John Lumby]

Thanks for the reply.   Yes, certainly.    I'll try to describe the problem 
I've faced numerous times with diff3.    It's the usual situation where I 
have made changes to version N and now want to merge them into version N+1.  
    Very often, the changes between N -> N+1 include minor reformatting 
changes, and sometimes I've also made such trivial changes in amongst the 
changes I've made.   I ran into just such a set recently.   Whenever a 
"significant" change of mine conflicts with a trivial reformatting change in 
N -> N+1, I want diff3 to ignore the trivial change and accept mine without 
flagging a conflict.    Similarly, where a trivial change of mine conflicts 
with a significant change in N -< N+1, I want to discard my trivial change.  
   That's what this -w option does (well, I think it does).

In practice I find that with this -w option, it reduces the number of 
conflicts quite dramatically.

Of course, this is not completely harmless - (some of) both my and the other 
guy's trivial changes are lost in the process.     However, this gives me 
the option to get something working much sooner.  Maybe later on I'll go 
back and reinstate the trivial changes, depending on what I'm doing it for.  
Moreover, very often these trivial changes are not put in by humans 
(manually) they are from some auto-indenter/tidier, and after I've finished 
getting my code working I can simply switch to whatever indenter I prefer or 
am asked to follow, and tell it to tidy the whole file.      I don't want to 
be confronted with any of that.

Here's an example from very recently of a fragment of emacs xfns.c that 
illustrates the problem.  If you put these three in files and try diff3'ing 
them with and without the -w you'll see what I mean.
I hope my and your e-mailer respects tabs - "mine" has no tabs, "older has 
leading tabs on lines 2, 3, 6 and 7  :  "yours" also has those tabs and tabs 
on lines 4 and 5 as well.    result - diff3 without -w sees a 3-way 
conflict.

John

mine:
#if 0 /* I don't know why it did this,
        but it seems logically wrong,
        and it causes trouble for MapNotify events.  */
     /* Match if the window is this frame's menubar.  */
     if (x->menubar_widget
         && wdesc == XtWindow (x->menubar_widget))
       return f;
#endif

older:
#if 0 /* I don't know why it did this,
         but it seems logically wrong,
         and it causes trouble for MapNotify events.  */
     /* Match if the window is this frame's menubar.  */
     if (x->menubar_widget
          && wdesc == XtWindow (x->menubar_widget))
        return f;
#endif

yours:
#if 0 /* I don't know why it did this,
         but it seems logically wrong,
         and it causes trouble for MapNotify events.  */
          /* Match if the window is this frame's menubar.  */
          if (x->menubar_widget
              && wdesc == XtWindow (x->menubar_widget))
            return f;
#endif




Original Message Follows----
From: Paul Eggert <address@hidden>
To: "John Lumby" <address@hidden>
CC: address@hidden
Subject: Re: extend diff3 to ignore whitespace differences - here's a patch
Date: 03 Dec 2003 15:19:12 -0800

Can you give an example of why this change is helpful?

What does it really mean for diff3 to ignore white space differences?
I.e., can you explain in general what diff3 should do, when given a
"diff" option that ignores some differences?  I'm looking for
documentation here, more than code.

_________________________________________________________________
Add photos to your messages with MSN 8. Get 2 months FREE*.  
http://join.msn.com/?page=dept/features&pgmarket=en-ca&RU=http%3a%2f%2fjoin.msn.com%2f%3fpage%3dmisc%2fspecialoffers%26pgmarket%3den-ca